No abstract

No abstract

No abstract

Vehicle re-identification (reID) plays an important role in the automatic analysis of the increasing urban surveillance videos, which has become a hot topic in recent years. However, it poses the critical but challenging problem that is caused by various viewpoints of vehicles, diversified illuminations and complicated environments. Till now, most existing vehicle reID approaches focus on learning metrics or ensemble to derive better representation, which are only take identity labels of vehicle into consideration. However, the attributes of vehicle that contain detailed descriptions are beneficial for training reID model. Hence, this paper proposes a novel Attribute-Guided Network (AGNet), which could learn global representation with the abundant attribute features in an end-to-end manner. Specially, an attributeguided module is proposed in AGNet to generate the attribute mask which could inversely guide to select discriminative features for category classification. Besides that, in our proposed AGNet, an attribute-based label smoothing (ALS) loss is presented to better train the reID model, which can strength the distinct ability of vehicle reID model to regularize AGNet model according to the attributes. Comprehensive experimental results clearly demonstrate that our method achieves excellent performance on both VehicleID dataset and VeRi-776 dataset.

In order to effectively aid the blind with optimal low-resolution vision and visual recovery training, pathfinding and recognition tests were performed using a simulated visual prosthetic scene. Simple and complex virtual scenes were built using 3DMAX and Unity, and pixelated to three different resolutions (32 × 32, 64 × 64, and 128 × 128) for real-time pixel processing. Twenty subjects were recruited to complete the pathfinding and object recognition tasks within the scene. The recognition accuracy and time required were recorded and analyzed after the trials. In the simple simulated prosthetic vision (SPV) scene, when the resolution was increased from 32 × 32 to 48 × 48, the object recognition time decreased from 92.19 ± 6.97 to 43.05 ± 6.08 s, and the recognition accuracy increased from 51.22 ± 8.53 to 85.52 ± 4.93%. Furthermore, the number of collisions decreased from 10.00 ± 2.31 to 3.00 ± 0.68. When the resolution was increased from 48 × 48 to 64 × 64, the object recognition time further decreased from 43.05 ± 6.08 to 19.46 ± 3.71 s, the recognition accuracy increased from 85.52 ± 4.93 to 96.89 ± 2.06%, and the number of collisions decreased from 3.00 ± 0.68 to 1.00 ± 0.29. In complex scenes, the time required to recognize the room type decreased from 115.00 ± 23.02 to 68.25 ± 17.23 s, and object recognition accuracy increased from 65.69 ± 9.61 to 80.42 ± 7.70% when the resolution increased from 48 × 48 to 64 × 64. When the resolution increased from 64 × 64 to 128 × 128, the time required to recognize the room type decreased from 68.25 ± 17.23 to 44.88 ± 9.94 s, and object recognition accuracy increased from 80.42 ± 7.71 to 85.69 ± 7.39%. Therefore, one can conclude that there are correlations between pathfinding and recognition. When the resolution increased, the time required for recognition decreased, the recognition accuracy increased, and the number of collisions decreased. Although the subjects could partially complete the recognition task at a resolution of 32 × 32, the recognition time was too long and recognition accuracy was not good enough to identify simple scenes. Complex scenes required a resolution of at least 48 × 48 for complete recognition. In addition, increasing the resolution shortened the time required to identify the type of room, and improved the recognition accuracy.

No abstract

No abstract

No abstract